Despite the non-convex nature of their loss functions, deep neural networks are known to generalize well when optimized with stochastic gradient descent (SGD). Recent work conjectures that SGD with proper configuration is able to find wide and flat local minima, which have been proposed to be associated with good generalization performance. In this paper, we observe that local minima of modern deep networks are more than being flat or sharp. Specifically, at a local minimum there exist many asymmetric directions such that the loss increases abruptly along one side, and slowly along the opposite side--we formally define such minima as asymmetric valleys. Under mild assumptions, we prove that for asymmetric valleys, a solution biased towards the flat side generalizes better than the exact minimizer. Further, we show that simply averaging the weights along the SGD trajectory gives rise to such biased solutions implicitly. This provides a theoretical explanation for the intriguing phenomenon observed by Izmailov et al. (2018). In addition, we empirically find that batch normalization (BN) appears to be a major cause for asymmetric valleys.

In this paper, we study a multi-step interactive recommendation problem, where the item recommended at current step may affect the quality of future recommendations. To address the problem, we develop a novel and effective approach, named CFRL, which seamlessly integrates the ideas of both collaborative filtering (CF) and reinforcement learning (RL). More specifically, we first model the recommender-user interactive recommendation problem as an agent-environment RL task, which is mathematically described by a Markov decision process (MDP). Further, to achieve collaborative recommendations for the entire user community, we propose a novel CF-based MDP by encoding the states of all users into a shared latent vector space. Finally, we propose an effective Q-network learning method to learn the agent's optimal policy based on the CF-based MDP. The capability of CFRL is demonstrated by comparing its performance against a variety of existing methods on real-world datasets.

One of the most common and studied problem in machine learning is classification. While conventional algorithms for supervised classification rely on the determination of a function from features to labels, we propose a different approach based on the estimation of a probabilistic transformation from features to labels. Indeed, we determine a conditional probability distribution of the labels given the features and then features are classified as labels following such distribution. In order to compute the conditional distribution, we follow a robust minimax approach, minimizing the worst-case expectation of the 0-1 loss. By doing so, we find the probabilistic transformation which achieves the minimum risk against an uncertainty set consistent with the training data. We show numerical results obtained by an implementation in python of this method and we compare its performance with state of the art techniques.

We propose a new numerical integration method for training a shallow neural network by using the ridgelet transform with a fast convergence guarantee. Given a training dataset, the ridgelet transform can provide the parameters of the neural network that attains the global optimum of the training problem. In other words, we can obtain the global minimizer of the training problem by numerically computing the ridgelet transform, instead of by numerically optimizing the so-called backpropagation training problem. We employed the kernel quadrature for the basis of the numerical integration, because it is known to converge faster, i.e. $O(1/p)$ with the hidden unit number $p$, than other random methods, i.e. $O(1/\sqrt{p})$, such as Monte Carlo integration methods. Originally, the kernel quadrature has been developed for the purpose of computing posterior means, where the measure is assumed to be a probability measure, and the final product is a single number. On the other hand, our problem is the computation of an integral transform, where the measure is generally a signed measure, and the final product is a function. In addition, the performance of kernel quadrature is sensitive to the selection of its kernel. In this paper, we develop a generalized kernel quadrature method with a fast convergence guarantee in a function norm that is applicable to signed measures, and propose a natural choice of kernels.

Existing graph- and hypergraph-based algorithms for document summarization represent the sentences of a corpus as the nodes of a graph or a hypergraph in which the edges represent relationships of lexical similarities between sentences. Each sentence of the corpus is then scored individually, using popular node ranking algorithms, and a summary is produced by extracting highly scored sentences. This approach fails to select a subset of jointly relevant sentences and it may produce redundant summaries that are missing important topics of the corpus. To alleviate this issue, a new hypergraph-based summarizer is proposed in this paper, in which each node is a sentence and each hyperedge is a theme, namely a group of sentences sharing a topic. Themes are weighted in terms of their prominence in the corpus and their relevance to a user-defined query. It is further shown that the problem of identifying a subset of sentences covering the relevant themes of the corpus is equivalent to that of finding a hypergraph transversal in our theme-based hypergraph. Two extensions of the notion of hypergraph transversal are proposed for the purpose of summarization, and polynomial time algorithms building on the theory of submodular functions are proposed for solving the associated discrete optimization problems. The worst-case time complexity of the proposed algorithms is squared in the number of terms, which makes it cheaper than the existing hypergraph-based methods. A thorough comparative analysis with related models on DUC benchmark datasets demonstrates the effectiveness of our approach, which outperforms existing graph- or hypergraph-based methods by at least 6% of ROUGE-SU4 score.

Russian President Vladimir Putin has ordered his government to put together a national strategy on artificial intelligence technology as Moscow is hustling to catch up on the research and development of the creation and implementation of intelligent machines, Defense One reports. "The Government of the Russian Federation, with the participation of Sberbank of Russia and other interested organizations, should develop approaches to the national strategy for the development of artificial intelligence and submit appropriate proposals" says an instruction sheet released by the Kremlin and approved by Putin. Their deadline is Feb. 25. China and the United States are far ahead of global competitors on the AI front, according to a report by the United Nations World Intellectual Property Organization that found U.S. firms IBM and Microsoft had the highest number of patents related to artificial intelligence. "The U.S. and China obviously have stolen a lead," WIPO Director-General Francis Gurry said during a news conference, per Reuters.

Since their entrance into mainstream political consciousness, Artificial Intelligence (AI) and Big Data have long been a harbinger of political doom and a coming onslaught of global, geopolitical shifts. Movies, TV series, think pieces and tech reports paint and increasingly grim picture of power being handed over by governments and citizens to amorphous algorithms that govern with no transparency. The most dramatic depiction is the all-out data-driven apocalypse of the Terminator universe, but subtler, more intimate insights into our Data Hell come from Black Mirror, whose episodes shed light on people, relationships and societies that have sacrificed their subjectivity in the name of optimization. In the political sphere, a mainstream position in the Democratic party of the U.S. is that Russia stole the 2016 Presidential Election with advanced hacking tools and trolls. The emerging race between the U.S. and China to develop the most advanced AI is being called the Cold War of the 21st century; the central power struggle that defines an era for the world.

China's internet firms are getting pally with giant state-owned automakers as they look to deploy their artificial intelligence and cloud computing services across traditional industries. Ride-hailing startup Didi Chuxing, which owns Uber China, announced earlier this week a new joint venture with state-owned BAIC. Hot on the heels came another entity set up between Tencent and the GAC Group. GAC, which is owned by the Guangzhou municipal government in southern China, announced Thursday in a filing it will jointly establish a mobility company with social media and gaming behemoth Tencent, Guangzhou Public Transport Group and other investors. The announcement followed an agreement between Tencent and GAC in 2017 to team up on internet-connected cars and smart driving, a deal that saw the carmaker tapping into Tencent's expertise in mobile payments, social networking, big data and cloud services.

AirAsia said it launched a chatbot named AVA, tapping artificial intelligence as it unveiled a redesigned mobile app and website. Svelte and dressed in the flaming red uniform of the low-cost carrier's flight attendants, AVA or AirAsia Virtual Allstar can answer frequently asked questions as well as provide flight information. The chatbot speaks 8 languages: English, Bahasa Malaysia, Thai, Bahasa Indonesia, Vietnamese, Korean, Simplified Chinese and Traditional Chinese, AirAsia said in a statement. Last October, AirAsia said it tapped Google Cloud to help integrate machine learning and artificial intelligence into its operations. AirAsia's mobile app has 3.3 million active users monthly, according to the statement.

Experts are warning of the threat posed by China's use of artificial intelligence (AI) to develop a survellience state, and say the risk of such authoritarian behaviour spreading to other parts of the world is increasing. While Chinese technology company Huawei is making daily headlines at the moment, Greg Austin, professor of cyber security, strategy and diplomacy at the University of New South Wales, said there were more pressing concerns. "If I were asked which was the bigger threat from China to the West, is it Huawei or is it their research on artificial intelligence I would say it's their research on artificial intelligence," Professor Austin said. "That far outranks any of the concerns that we have from what Huawei might do in terms of foreign espionage." Huawei has been banned from taking part in the rollout of 5G mobile technology in Australia over national security concerns and has faced similar restrictions in other countries.